Introduction

The agricultural sector represents a significant source of greenhouse gas (GHG) worldwide due to direct and indirect emissions of carbon dioxide (CO.), methane (CH.) and nitrous oxide (N.O). Estimates have shown that agriculture contributes to the enhanced GHG effect by emitting around 7.1 Gt of CO.eq, or ∼18% of total global anthropogenic GHG emissions (Gerber et al., 2013).

Agricultural sources are responsible for 38.09% of Colombia's GHG total emission (62 million t of CO. per year). The main partition of emission of GHG are 49.8% for CO., 30.1°% for CH. and 19.1°% for N.O, being those two last gases mainly related to agricultural sources (Pedraza et al., 2009).

Worldwide, the systems that are exposed to intensive uses has higher utilization of agricultural inputs as soluble fertilizers, mainly nitrogen and pesticides (Smith et al., 1997; Lal, 2004; Tubiello et al., 2013), practices that results in direct and indirect GHG emissions (Lal, 2004).

Pesticide manufacturing represents about 9% of the energy used for arable crops (IPCC, 2006; Lal, 2004). It is assumed that due course all the carbon included in the pesticide will be broken down and emitted to the atmosphere as carbon dioxide (IPCC, 2006).

Agroforestry systems can have a major effect on the productivity of fruit production systems, where nitrogen is often a limiting factor in production systems (Nair et al., 2009). Agroforestry systems can substantially reduce the use of synthetic fertilizers through the biological nitrogen fixation (BNF) of leguminous tress (Nair et al., 2009; Naranjo et al., 2012).

Agroforestry is based on principles of sustainable production and diversification (Nair et al., 2009;Naranjo et al., 2012); an agroforestry system retains soil and biomass carbon stocks (Lal, 2011). Moreover, the soil carbon stock is related to management soil factors as land use, residue inputs of soil C and soil tillage practices (Albrecht and Kandji, 2003; IPCC, 2006).

Lal (2011) estimated that 89% of the agriculture sector's total GHG mitigation potential is from soil organic carbon (SOC) sequestration. Carbon sequestration has the potential to offset fossil fuel emissions by 0.4 to 1.2 gigatons of carbon per year (Lal, 2011).

The objective of this study is to estimate the GHG balance related to conventional and sustainable fruit production systems in the Highlands region of Pasto, Colombia, in order to identify the fruit production system that emits less GHG emissions and has greater potential to mitigation GHG emissions (soil and biomass C sequestration).

Materials and methods

Location and production systems

The systems considered in our study refer to the fruit production located in the highlands region of Pasto, state of Nariño, South-West Colombia. Geographical coordinates of 0°37' to 2°47'N and 79°03' to 76°47'W. It is one of the highest plateaus of the country located between 2,400 to 2,800 m a.s.l.

Three types of conventional fruit production systems were found in the various degraded areas of the Highlands region of Pasto, Colombia, were as follows:

Monoculture of Rubus glaucus Benth. (S1); monoculture of Physalis peruviana (S2) and monoculture of Solanum quitoense Lam. (S3). The cultivation of these systems had used the conventional type. On the other hand, three types of sustainable fruit production systems were found in the various areas of the highlands region of Pasto, were as follows: Agroforestry system of Rubus glaucus Benth. (SS1); agroforestry system of Physalis peruviana (SS2); agroforestry system of Solanum quitoense Lam (SS3). The cultivation of these systems had been associated with Acacia decurrens trees as living fence and live coverage of kikuyu Pennisetum clandestinum grass. In this study, data was collected from 20 actual farms for each one of the systems considered. Tab. 1 presents the details associated with the agronomic parameters of the fruit production systems considered in this study. All of the systems were run in a 06-year cycle (Tab. 1).

Characterization of production systems

Emission sources and sinks and amount of supplies

Table 2 presents the sequence of sources and potential sinks related to the main greenhouse gases (CO., CH. and N.O) associated to each of the production systems under analysis in this study.

Table 3 and Figure 1 present the agricultural supplies and fuel consumption due to agricultural activities conducted in each of the studied systems corresponding to variables from 1 to 8 and analysis de continuous variables in multivariate analysis.

Emission factors

N.O from soil management and CO. from agricultural activities

The direct plus indirect emissions from N fertilizer applications and above ground residues were estimated by using the IPCC (2006) methodology. Emission factor regarding lime was assumed as 0.477 kg CO.eq kg^-1 (dolomite) (IPCC, 2006). Emission factors associated with the manufacturing, transport and storage of potassium and phosphate fertilizers were 0.2 kg CO.eq kg^-1 for P and 0.15 kg CO.eq kg^-1 for K, as proposed by Lal (2004). For pesticides, the emission factor EF depends on the type of pesticide applied (Helsel, 1992) to control pests and diseases in all systems.

An emission factor considered as 2.33 kg CO.eq L^-1 of gasoline, under tropical conditions (IPCC, 2006).

Greenhouse gas emissions variables were expressed in CO. equivalent units to account for global warming potential of each gas in accordance with IPCC (2006), assuming a 100-year time horizon (298 for N.O and 1 for CO.) (Table 4, variables from 9 to 19).

Soil and biomass C pools

The estimative of potential sinks either in soil or in biomass is presented in Tab. 5 (kg CO.eq ha^-1 yr^-1) considering the 6-year cycle.

Reference values for the soil C stock in conventional S1, S2 and S3 fruit production systems of Highlands region of Pasto, Colombia, was based on soil analysis, being these 156.76, 51.84, 169.8 t C ha^-1 in the top 30 cm layer. These values were used for estimating changes in final soil carbon stocks (V21) (Tab. 5) by converting from conventional to sustainable fruit production systems.

Ratios of gains/losses of soil C (V22) (Tab. 5) in the studied fruit production systems were estimated by using specific methodology proposed by IPCC (2006), which takes into account factors related to soil management practices: land use (FLU), tillage practices (FMG) and residue inputs (FI) for a time-period of 20 years (IPCC, 2006). In addition to the intensity of management adopted (for instance, high, medium and low inputs) those factors take into account climate and soil type in the specific region. Rates of gains/ losses of soil C were multiplied by 3.66 to convert it from C to CO.eq (V23) (Tab. 5).

Accumulation rate of C in biomass was estimated only in sustainable fruit production system. The increase in biomass C stock was assumed as 4.5 t C ha^-1 year^-1, considering the wood component, based on the IPCC (2006) methodology for Acacia ssp in South America. The accumulation rate of C in biomass of Pennisetum clandestinum grass of 0.16 t C ha^-1 year^-1 was calculated of Giraldo et al. (2008); these rates were expressed in kgCO.eq and related to (V24) variable (Tab. 5).

Total GHG balance

The results of total GHG balance emission (V25) considering the potential for soil C gain/loss and biomass C fixation are reported on Tab. 5.

Statistical analysis

A Principal Components Multivariate Analysis was performed in order to reduce the number of explanatory variables, using variables that were not collinear. A numerical classification of farms was then performed using cluster analysis method with the same variables identificated. All analyses were conducted with SAS. software.

Results and discussion

Correlation between variables

The correlation matrix description of emission sources and total GHG emissions variables under different fruit production systems of Highlands region of Pasto, Colombia indicated that the variables (V12) (CO. from agricultural activities), (V20) (total GHG emissions) and (V25) (GHG balance emissions) considered independently, demonstrated positive correlation with others variables. The (V12) variable (CO. from agricultural activities) was best explained by the (V4) variable (lime) (r=0.95), the (V13) variable (CO. from lime) (r=0.95) and the (V14) variable (CO2 from P) (r=0.90).

According to the Intergovernmental Panel on Climate Change IPCC (2006), CO . emissions from all lime added in the year of application although the effect of liming usually lasts for a few years (after the new addition of lime), depending on climate, soil and cultivation practices (IPCC, 2006). For instance, all C in lime is eventually released as CO. to the atmosphere (IPCC, 2006). Emission factors of phosphates and potassic fertilizers are associated with manufacturing, transportation, storage and application. On agroforestry systems, nutrient recycling is higher, reducing dependence on lime, phosphate and potassic fertilizers (Nair et al., 2009).

On the other hand, the (V20) variable (total GHG emissions) was increased when increased the (V3) variable (N from crop residues) (r=-0.89) and the (V11) variable due to N.O from N crop residues emissions (r=-0.88). Soil C losses in terms of CO. emissions can be as high as the annual C sequestration rates due to N.O from N crop residues emissions occasioned by conventional tillage (La Scala etal., 2008). N.O is a gaseous by-product of nitrification that is ultimately released into the atmosphere (IPCC, 2006).

The variables that were highly correlated with the (V25) variable (total GHG balance emissions) were the (V22) variable (rates of gains/losses soil C) (r=-0.99) and the (V24) variable (biomass C fixation) (r=0.95).

Soil carbon sequestration is a process in which CO. is removed from the atmosphere and stored in the soil carbon pool, primarily mediated by plants through photosynthesis (Lal, 2011). Sustainable fruit production systems showed a large potential of sequestering carbon in soil and biomass, as observed by Giraldo et al. (2008) in an agroforestry system located in an Andean region of Colombia; which suggests the importance of the agroforestry fruit systems evaluated through on GHG mitigation.

Multivariate analyses

To make a distinction between the systems analyzed, principal components were generated (Factor 1 and Factor 2). The PCA considered the first two factors with a cumulative value of 66.84 % for the variables analyzed (Table 6) was negatively associated with the variables V1 (N synthetic fertilizers) (r=-0.93), V4 (lime) (r=-0.97), V9 (N.O from N synthetic fertilizers) (r=-0.93), V13 (CO. from Lime) (r=-0.98), V18 (CO. from agricultural activities) (r=-0.93), and the V20 (total GHG emissions) (r=-0.95), being the most sensitive variables in these analyses (Tab. 6).

However, these variables can also be observed in the vector diagram, where the variables are closer to the axis of this factor to demonstrate that most can influence the distinction between the types of fruit systems evaluated (Fig. 3). Diagram generated for the projection vectors demonstrated that the (V22) variable (rate soil C gains/losses) (r=0.85) was that most positivity influence the distinction between the types of fruit production systems (Fig. 3).

Despite the huge potential for mitigation of GHG emissions, especially in sustainable fruit production systems, it is important to point that soil C accumulation could be lost rapidly, depending on the soil management decisions made at those sites. For instance, Conant et al. (2001) reviewed about 115 studies in 17 countries on the effects by conversion from agricultural crops to agroforestry system on soil C accumulation. This author considered values of soil C sequestration rates ranged from -0.2 to 3.0 t C ha^-1yr^-1 respectively.

It could be concluded that better soil management is possible through the use of agroforestry systems as demonstrated also by Nair et al. (2009) and Giraldo et al. (2008).

For factor 2, the variables V2 (N organic fertilizer) (r=0.91) and V21 (Soil final C stock) (r=0.86) explained 16.9 % of the variation (Tab. 6). The increase in soil C stock is subject to greater amounts of crop residues returned to the soil (Albrecht and Kandji, 2003) and minimal soil disturbance (Johnson et al., 2010) (Tab. 6).

The cluster analysis performed with the same variables as the principal components analysis identified three groups (Fig. 4).

The cluster analysis showed that the first cluster consisted of conventional (S3) and (S2) fruit systems, with no statistical differences; in the (S3) system, characterization with continuous variables of cluster or categories showed that the variables that had greater weight was total GHG balance emissions (V25) (4,407 kg CO.eq ha^-1 yr^-1) (P= 0.046) (Fig. 4), due to that has on its favor the highest soil C losses (V23) (P= 0.048) (7,523 kg CO.eq ha^-1 yr.) and an additional potential for GHG emissions (V20) (P= 0.037) equivalent to 1,470 kg CO.eq ha^-1 yr^-1, statistically equating to (S2) system (Figure 4) with total GHG balance emissions (V25) of 5,544 kg CO.eq ha^-1 yr^-1, further emissions would be expected according to our estimations due also to potential of soil C losses (V23) and total GHG emissions (4,404 and 1,104 kg CO.eq ha^-1 yr^-1) (V20).

The multivariate analyses also showed that N from the synthetic fertilizers (V10) significantly influenced the formation of this cluster (P= 0.029). The use of N synthetic fertilizers in agriculture in Colombia is 137 kg ha^-1, almost double the intensity in South American, with an average of 74 kg ha^-1 (World Bank, 2008), resulting in higher direct and indirect N.O emissions (Smith et al, 1997).

The formation of the cluster 2 as showed in Fig. 5, agroforestry system of Rubus glaucus Benth. (SS1) can neutralize higher emissions (-21,079 kg CO.eq ha^-1 yr^-1) (V25), as according to our results has, in addition to the higher potential soil C accumulation (-4,587 kg CO.eq ha^-1 yr^-1) (V23) and lower total GHG emissions (610 kg CO.eq ha^-1 yr^-1) (V20), results from this study are compared with relevant studies of Naranjo et al. (2012). It was statistically similar to agroforestry system of Physalis peruviana (SS2) (Fig. 4) accounting for total GHG balance emission reduction of -20,846 kg CO.eq ha^-1 yr^-1 (V25).

But in turn the agroforestry system of Physalis peruviana (SS2) was statistically equals to monoculture of Rubus glaucusBenth. (S1) (Fig. 4), although it is a system that does not neutralize GHG emissions but if it emits less GHG to the atmosphere (4,407 kg CO.eq ha^-1 yr^-1) than the other two conventional (S2) and (S3) fruit production systems.

The cluster analysis in Fig. 4 showed an intermediate cluster to the agroforestry system of Solanum quitoense Lam. (SS3) system, accounting for a total GHG balance emissions reduction of -19,519 kg CO.eq ha^-1 yr^-1 (V25); statistically it may be equal to sustainable fruit production (SS1) and (SS2) systems, but also to the conventional fruit production (S1) system.

It is important to point that differences in management practices by adoption from conventional (S1) monoculture of Rubus glaucus Benth. to sustainable (SS1) agroforestry system of Rubus glaucusBenth. could significantly affect subsequent trends in increases of soil carbon accumulation and potential for mitigation of the GHG emissions.

Conclusion

Sustenaible fruit production systems (agroforestry) have the potential to offset GHG emissions, representing an important alternative to the recovery of degraded areas of conventional fruit production systems in Highlands region of Pasto, Colombia because they are able to maintain biomass C and soil organic matter through the addition of litter and crop residues in the soil.

Literature cited

Albrecht, A. and S.T. Kandji. 2003. Carbon sequestration in tropical agroforestry systems. Agric. Ecosyst. Environ. 99, 15-27. Doi: 10.1016/S0167-8809(03)00138-5.

Angulo, R. 2006. Lulo el cultivo: Solanum quitoense Lam. Fundación Universidad de Bogotá, Jorge Tadeo Lozano, Bogotá.

Angulo, R. 2011. Uchuva Physalis peruviana. Bayer Crop Science, Bogotá.

Conant, R.T., K. Paustian, and E.T. Elliott. 2001. Grassland management and conversion into grassland: effects on soil carbon. Ecol. Appl. 11, 343-355. Doi: 10.1890/1051-0761(2001)011[0343:GM ACIG]2.0.CO;2.

Pedraza G., A., M. Cabrera L., M. Duarte O., M.M. Gutiérrez A., P.S. Lamprea Q., and R.J. Lozano P. 2009. Visión general del inventario nacional de fuentes y sumideros de gases de efecto invernadero. pp. 14-50. In: Instituto de Hidrología, Meteorología y Estudios Ambientales - IDEAM. Inventario nacional de fuentes y sumideros de gases de efecto invernadero 2000 - 2004. Bogotá.

Gerber P.J., H. Steinfeld, B. Henderson, A. Mottet, C. Opio, J. Dijkman, A. Falcucci, and G. Tempio. 2013. Tackling climate change through livestock - A global assessment of emissions and mitigation opportunities. FAO, Rome.

Giraldo, A., M. Zapata, and E. Montoya. 2008. Carbon capture and flow in a silvopastoral system of the Colombian Andean zone. Arch. Latinoam. Prod. Anim. 16, 241-245.

Giraldo, L., J. Botero., J.Y. Saldarriaga, and P. David. 1995. Efecto de tres densidades de árboles en el potencial forrajero de un sistema silvopastoril natural en la región Atlántica de Colombia. Rev. Agrof. Amer. 2, 14-19.

Helsel, Z.R. 1992. Energy and alternatives for fertilizer and pesticide use. pp. 177-201. In: Fluck, R.C. (ed.). Energy in farm production. Elsevier, Amsterdam. Doi: 10.1016/B978-0-444-88681-1.50018-1.

IPCC, Intergovernmental Panel on Climate Change. 2006. IPCC 2006 Guidelines for national greenhouse gas inventories. In: Eggleston, H.S., L. Buendía, K. Miwa, T. Ngara, and K. Tanabe (eds.). Agriculture, forestry and other land use. Nacional Greenhouse Gas Inventories Programme; Institute for Global Environmental Strategies (IGES), Hayama, Japan.

Johnson, M., R. Edwards, and O. Masera. 2010. Improved stove programs need robust methods to estimate carbon offsets. Clim. Change 102, 641-649. Doi: 10.1007/s10584-010-9802-0.

Lal, R. 2004. Carbon emission from farm operations. Environ. Intl. 30, 981-990. Doi: 10.1016/j.envint.2004.03.005.

Lal, R. 2011. Sequestering carbon in soils of agro-ecosystems. Food Policy 36, 533-539. Doi: 10.1016/j.foodpol.2010.12.001

La Scala, A., K. Lopes, D. Bolonhezi, D.W. Archer, and D.C. Reicosky. 2008. Short-temporal changes of soil carbon losses after tillage described by a first-order decay model. Soil Till. 99, 108-118. Doi: 10.1016/j.still.2008.01.006.

Nair, P.K., B.M. Kumar, and V.D Nair. 2009. Agroforestry as a strategy for carbon sequestration. J. Plant Nutr. Soil Sci. 172, 10-23. Doi: 10.1002/jpln.200800030.

Naranjo, J.F., C.A. Cuartas, E. Murgueitio, J. Chará, and R. Baraho-na. 2012. Balance de gases de efecto invernadero en sistemas silvopastoriles intensivos con Leucaena leucocephala en Colombia. Livest. Res. Rural Dev. 24, 1-12.

Silva, A., C. Bucheli, A. Castillo, O. Checa, and T.L. Lagos. 2015. Respuesta de Physalis peruviana a la fertilización con diferentes dosis de N, P y K en el Altiplano de Pasto. Acta Agron. 64, 330-335. Doi: 10.15446/acag.v64n4.44290.

Smith, K.A., I.P. Taggart, and H. Tsuruta. 1997. Emissions of N.O and NO associated with nitrogen fertilization in intensive agriculture, and the potential for mitigation. Soil Use Manage. 13, 297-304. Doi: 10.1111/j.1475-2743.1997.tb00601.x.

Tubiello, F.N., M. Salvatore, S. Rossi, A. Ferrara, N. Fitton, and P. Smith. 2013. The FAOSTAT database of greenhouse gas emissions from agriculture. Environ. Res. Lett. 8, 015009. Doi: 10.1088/1748-9326/8/1/015009.

World Bank. 2008. Colombia, Costa Rica and Nicaragua: integrated silvopastoral approaches to ecosystem management project. Washington, DC.